FASTA and BLAST

1
FASTA and BLAST

• Chitta Baral

2
FASTA Basic Steps

• Step 1
• Set a word size. (usually 6 for DNA and 2 for
  proteins)
• Make a plot.
• Find the long diagonals (or high scoring regions)
• Step 2
• Score the 10 best diagonal runs using a scoring
  matrix. (allow mismatches, end extensions,
  joining of two diagonals but no gaps)
• (init1 single best sub-alignment found in this
  stage.)
• Step 3
• Merge non-overlapping diagonal runs to allow gaps
  (ins/del).
• Score of joined regions sum of individual
  scores penalty
• Score of the highest scoring region at the end of
  this step is called initn.
• Step 4
• Use a variant of Smith-Waterman algorithm on a
  narrow band around initn and construct an optimal
  alignment of this region.
• Modifications
• In Step 4, use a band around init1.

3
BLAST basic steps

• Step 1
• Set a word size (3 for protein and 11 for DNA)
  Create a word list for the query sequence
• Eg. qlnfsagw ? ql, ln, nf, fs, sa, ag, gw
• Expand the list (using a threshold T, say 8)
• ql ql, qm, hl, zl ln ln, lb
• nf nf, af, ny,df,qf, ef, gf, hf, kf, sf, tf, bf,
  zf fs fs, fa fn, fd, fg, fp, ft, fb, ys
• sa none ag ag
• gw gw, aw, rw, nw, dw, qw, ew, hw, iw, kw, mw,
  pw, sw, tw, vw, bw, zw, xw
• Step 2
• Scan through the string and whenever a word in
  the list is found try to extend it in both
  directions (no gaps) to get to a score beyond a
  threshold S. While extending use a parameter L
  that defines how long an extension will be tried
  to raise the score over S.
• Modifications of Step 2
• Original BLAST extension is continued as long as
  the score continues to increase
• Another version extension is stopped when the
  accumulated score stops increasing and has just
  begun to fall a certain amount below the best
  score found.
• Blast2 (gapped BLAST)
• Lower value of T is used
• After extension try to combine (allowing gaps)
• Find maximal scoring segment. Use Smith-Waterman
  algorithm around a band of this segment (as in
  FASTA)

4
Home Work (due 3/31/03)

• Compare BLAST and FASTA.
• (Hint Read the external pointers in the class
  notes page.)